After high nutrient inputs stopped, plant composition was closer to controls mainly with annual fire in Cedar Creek field B, 2015–2022

We tested whether plant composition returned toward untreated controls after 32 years of nutrient addition stopped. We used Cedar Creek experiment E002, field B, in Minnesota, USA. The main sample had 18 plots at 9.52 to 27.2 g N per square metre per year, with six sample years from 2015 to 2022. Each plot had a half with continued nutrient input and a half where input stopped after 2013. Nine plots had annual spring fire since 1992. We measured Bray-Curtis distance between species biomass shares and six untreated controls in the same year. Stopped halves were closer to controls by 0.156 distance units with annual fire and by 0.020 without annual fire. The fire difference was 0.136 (95% bootstrap interval 0.087 to 0.186; exact fire test p = 0.00025, 8000 assignments). The fire difference rose from 0.020 in 2015-2016 to 0.255 in 2021-2022. It held with fixed controls across years and with presence-absence data. Agropyron repens and Schizachyrium scoparium gave most of the distance change. Recovery was incomplete: burned stopped halves had a mean distance of 0.795, while distinct controls had a mean distance of 0.329. This extends a prior richness analysis of the same data. Limits are one field, 18 main plots, small strips, and no confirmed random half assignment for field B. All added nutrients stopped together. The result concerns a long fire regime and does not test fire started at cessation. Code, source data, provenance, figures, and independent numerical checks are preserved.

grassland recovery · plant composition · nutrient cessation · annual fire · Cedar Creek · Bray-Curtis · reference uncertainty · reproducible ecology